Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/jazbengu/ThreatDetect/llms.txt

Use this file to discover all available pages before exploring further.

ThreatDetect is an AI-powered insider threat detection platform that analyzes employee behavioral data and flags individuals who exhibit malicious patterns. It combines an XGBoost classifier with Isolation Forest anomaly detection, then uses SHAP values to explain every prediction in plain language — so your security team understands not just who is flagged, but why. This documentation covers everything you need to install ThreatDetect, prepare your data, run analyses, and interpret results.

Quickstart

Get ThreatDetect running and analyze your first dataset in minutes.

How it works

Understand the detection pipeline: features, model, scoring, and explainability.

CSV batch analysis

Upload an organizational CSV and detect threats across your entire workforce.

Input data schema

Learn exactly which columns your data must contain for accurate predictions.

Key features

  • Ensemble detection — XGBoost classifier combined with Isolation Forest anomaly scoring produces robust risk predictions that reduce false positives.
  • SHAP explainability — Every prediction, individual or organizational, is accompanied by SHAP-based feature contributions that show which behavioral signals drove the result.
  • Three analysis modes — Run a batch scan across your entire organization via CSV upload, analyze a single employee interactively, or explore your dataset visually with the built-in EDA page.
  • Downloadable results — Batch analysis produces a results CSV with Prediction, Risk_Prob, Anomaly_Score, and Confidence columns for every employee record.
  • Global and per-instance explanations — View organization-wide feature importance charts alongside SHAP waterfall plots for individual employees.
ThreatDetect relies on a pre-trained model file located at AI_Model_Code/insider_threat_model.pkl. This file must be present in the repository root before you launch the app. See the installation guide for details.

Build docs developers (and LLMs) love